Ian D. said:
I have never really bothered to learn much about PSUs. I've always
bought the expensive ones and that strategy has payed off for 10
years.
Until now.
So: Is it the responsibility of the PSU to deliver the 3.3V? I think
it is, yes?
So, If I have 4V instead, the PSU should be faulty, right?
The PSU feature protective circuitry and shuts down to protect
computer components, right?
So, If I have Bios and MB stutdowns disabled, and the computer just
suddenly instantly shuts down black, it's likely to be the PSU that
shuts down, right? (monitors 3.3V alarm squeeking, = 4V)
It depends on the era of motherboard. Some motherboards used
to make something called "VIO", back when memory ran at
3.3V. The motherboard might have an option to boost the VIO
voltage to 3.5 or 3.6V. But the VIO was likely not monitored
by the hardware monitor chip on those motherboards, and
the monitor chip was connected to the real 3.3V coming from
the PSU. (The 3.3V might be used by PCI cards, for example.)
Yes, it is the responsibility of the PSU to deliver 3.3V.
You will find at least one pin on the ATX power connector,
that has two wires going to it. One of the wires on that
pin, is a remote sense, intended to allow the PSU to more
closely monitor the output of 3.3V. In theory, that should
allow it to be regulated more closely (for example, nulling
out the voltage drop in the cable).
PSUs can have a variety of features, and a feature rich
PSU might list overvoltage and overcurrent protection. But
the overcurrent is hard to fit to the cheapest way of
designing a PSU, and the current values listed suggest the
power supply is detecting an overload thermally, instead
of using a precise electronic method.
Overvoltage is a bit easier to do. Since overvoltage happens
so seldom, I cannot really tell you if your experience is
typical or not.
One thing to consider about a PSU, is how well it is able
to protect itself. What I'm referring to here, is whether a
PSU is "push-pull" or only "push". Say, for example, I connect
a low value resistor from the 5V rail, to the 3.3V rail.
A power circuit that was "push-pull", would simply sink
the extra current coming from the 5V rail, and the output
would remain at a steady 3.3V. If the supply could only
"push" current, a +5V to +3.3V fault would result in the
power supply stopping the pushing of current, but the
voltage on the output would continue to rise. Such a power
supply then, has no way to defend itself from a rail to
rail fault. And the overvoltage circuit, if one is present,
is the only opportunity for the fault to be stopped (by
shutting down all outputs).
Another condition for power supplies, is their minimum
load requirement. Some supplies list 0 amps as the minimum
current. Other supplies might list 2 amps or so as the
minimum. What that means, is the output is not tightly
regulated, if the load doesn't manage to meet the minimum
value. So the voltage might go mildly off from the
desired value, or the voltage could exceed the +/- 5%
or so value, printed on the label on the side of the supply.
Generally, you would not expect the design to rise to
dangerously high values, if no load was present. But you
wouldn't go out of your way to encourage the supply to
do it either.
So, where does that leave you ?
To test the supply, disconnect it from the motherboard.
Apply a minimum load to it. I use maybe $30 worth of
load resistors to do that. Then I connect a voltmeter
to the outputs, while the load is in place, and
verify my PSU voltages. Having a minimal load on the
supply, helps to ensure the regulation is going to work.
Even without that care and attention, if it looks like
you'll be throwing out the supply anyway, try running
it without a load, connect PS_ON# to COM to turn it
on, then measure the voltages. It is possible it will
measure normal in this condition. If it does measure
normal, it could be that your motherboard has a
phantom connection between 3.3V and 5V or 3.3V and 12V.
If a supply voltage more positive than 3.3V, leaks
current into the 3.3V plane on the motherboard, then
the voltage could climb to 4V. If the supply has OVP,
then the supply could shut itself off, in response to
seeing the voltage so high.
So you still don't know for certain yet, where the
fault lies.
As for the motherboard end of things, I don't know if
all the BIOS out there, would see a high value of
voltage on a rail and shut down. I would expect
beeping or some other sound effect, but I don't know
if a motherboard would shut down as a response.
I just checked one power supply spec, and on it,
the OVP was set to 4.1V on the 3.3V rail. So that
is roughly consistent with the power supply
initiating the shutdown. If you can find a spec
for the PSU, it might list the OVP value for
the design.
There is a real supply schematic here, if you want
to see how an older supply works:
http://www.pavouk.comp.cz/hw/en_atxps.html
HTH,
Paul
